glean

Glean MCP server integrates the Glean API to provide two capabilities: a Search function that returns a list of relevant results for a given query, and a Chat function that enables Q&A with a built-in chatbot using the Glean context. Once deployed, you can access these tools through your MCP-enabled interface (for example, via Claude Desktop) and issue queries to retrieve search results or engage in conversational Q&A. The server is designed to run as a Docker container named glean-server and relies on two environment variables to authenticate and target the Glean instance.

To use the tools, configure your MCP integration with the glean-server entry and provide the required environment variables GLEAN_API_KEY and GLEAN_DOMAIN. The Search tool will return a structured list of results for a user’s query, while the Chat tool facilitates interactive dialog with the Glean-powered chatbot using the same underlying context.

Prerequisites:

• Docker installed on your machine (Linux, macOS, or Windows with Docker Desktop).
• Access to a Glean API key and your Glean domain.

Installation steps:

1. Build the MCP server Docker image used by this integration:

docker build -t glean-server:latest -f src/glean/Dockerfile .

2. Run the server container (example command shown in the MCP config):

docker run -i --rm \
  -e GLEAN_API_KEY=YOUR_API_KEY_HERE \
  -e GLEAN_DOMAIN=YOUR_DOMAIN_HERE \
  glean-server

3. Add the server to Claude Desktop (or your MCP client) configuration, using the snippet from the README:

{
  "mcpServers": {
    "glean-server": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "GLEAN_API_KEY",
        "-e",
        "GLEAN_DOMAIN",
        "glean-server"
      ],
      "env": {
        "GLEAN_API_KEY": "YOUR_API_KEY_HERE",
        "GLEAN_DOMAIN": "YOUR_DOMAIN_HERE"
      }
    }
  }
}

4. Verify connectivity by issuing a test query through the MCP interface to ensure the Search and Chat tools respond as expected.

Tips and common issues:

• Ensure Docker is running and you have network access to the Glean API endpoint specified by GLEAN_DOMAIN.
• Keep GLEAN_API_KEY confidential; rotate keys if needed and update the environment variable accordingly.
• If the container fails to start, check Docker logs for errors related to missing dependencies or incorrect environment variables.
• The image name used in the config is glean-server; ensure the tag matches (latest in the build command).
• You can customize the environment by adding extra variables supported by your Glean setup if needed (not required for basic operation).
• If you hit rate limits, consider caching frequent queries at the MCP level or using domain-specific configurations provided by Glean.